| From | Message | ||
|---|---|---|---|
|
vermonty 29-Nov-10, 14:29 |
 Rating system bugslow. Today I won my sixth rateable game; it was against a player who has an established rating about 30 points lower than my first rating. Now my new rating is only *one* point higher than my first rating. What is going on here?!? Does GameKnot's rating system have a feature that artificially deflates strong performances by new players? Will this bug be fixed? I don't care about my rating. I care about the integrity of the rating system here. GameKnot management, please respond to my concern! Thanks. |
||
|
According to GK's own explanation => gameknot.com your provisional rating should be ( (1490 + 400) + (1593 + 400) + 1845 + (1682 + 400) + (1572 + 400) + (1767 + 400) ) / 6 = 1992 so I don't understand much of it either |
||
|
vermonty 06-Dec-10, 21:25 |
GameKnot rating system is still partly broken that they publish. I was rated at 1797 based on five games against established opponents, and then I defeated an established 1767 and gained just one (1) point for that win. Will GameKnot fix these errors? Will they repair their broken rating system? Will they reply to the message that I sent to customer service on 30-November? All I've received so far is a meaningless form letter dated 2-December. I repeat, GameKnot is not following its published rules for rating new players! |
||
|
baronderkilt 06-Dec-10, 22:39 |
one point? |
||
|
|
||
|
vermonty 07-Dec-10, 10:11 |
Deleted by vermonty on 09-Dec-10, 13:13.
|
||
|
Now that we had a chance to look into this issue, it turns out that due to technical reasons one's rating cannot change more than +/-127 points after a single game (the more computer oriented among you will probably recognize it as a signed byte). Obviously this doesn't affect any players with an established rating, considering that because of the Elo formula the theoretical maximum is +/-20 points starting with the 20th game anyway. It does mean that after the first 5 games, which are probably the only ones that can be affected, the maximum rating a player can reach is 1835 (1200+5*127). After scanning the past games histories, it appears only about a dozen of players where affected by this in the past year, and vermonty was one of them. Thank you for reporting the problem, and please accept our sincere apologies that we've have had this unpublished limitation in our rating system. We'll look into ways of fixing it, but it probably won't be possible to make it retroactive (we would have to recalculate everyone's ratings, as they all are relative to each other). The good news is, the Elo rating system works in such a way, that if your true rating is higher than your assigned rating, you will be getting more points for winning games and losing less -- it is self-correcting, which is one of the reasons why it works so well. Simply play and win more games, and your rating will be reflecting your true chess-playing skills in no time. |
||
|
|
||
|
vermonty 07-Dec-10, 18:39 |
Thank you, GameKnot, for your good explanation excellent. Thank you. There is no need for you to change my rating now. During my first few games on GameKnot, I was trying much harder than my opponents. Now that I'm playing team games, I see that my current rating is fair and reasonable. I'm not as strong as I was earlier, and my opponents are playing better. |
||
|
baronderkilt 08-Dec-10, 08:10 |
Yes ... vermonty *** I really just wanted to mention something to you about Team Play ratings. As FYI. It seems they usually come out to be somewhat lower for most players than their general GK rating. For which I think there are 2 factors involved. (SOME do manage to be higher Team Rated, but that's the "exception" ) 1) Most players do seem to put more effort or at least Resistance into their Team Games. Many have said this. No one wants to hurt their team! ...and want to play their best there. 2) Secondly, I believe the Team Ratings of players are calculated as a separate "Rating Pool" from the overall GK Ratings. For instance, if you win a team game it is calculated to add to your Team Rating and also calculated into your GK Rating; as 2 separate rating groups. IMO the Team Rating group/pool seems to be Lower rated overall. An example would be to compare Fischer's 70's rating of 2700-ish FIDE, but 2800+ USCF just because they were 2 separate rating groups, calculated separately. And the USCF ratings for everyone tended to approach 100 pts more than FIDE ratings of the time. (Largely due to the "Bonus Points" Elo that USCF was then using, that FIDE did not use. IMO) I believe this is analogous to the GK situation of Team Rating vs GK Rating. Perhaps GK will correct me if this is wrong . *** The other thing I wanted so say is, Congratulations~! .... on your fine start here and rating. Well done. Regards. Craig A.C. |
||
|
vermonty 08-Dec-10, 20:34 |
I agree, the that the "Team" [team play, really] ratings are the ones that are much too low. I would like to see the "Team" ratings abolished. I don't think they serve a useful purpose, and they are confusing to new players. Alternatively, GK management could make a one-time adjustment to bring the average "Team" rating to parity with the average "Player" rating. I would like to see "Team rating" used to describe the average rating of an entire team or a substantial subset of an entire team. For example, one could take the average "Player rating" of all members of the team who have been active within the past 90 days or six months. |
||
|
vermonty 08-Dec-10, 20:38 |
There's a forum bug! I agree, the "Team" ratings are too low ... But it got truncated. Perhaps the use of a quotation mark triggers the end of the Subject line. FYI, the subject line of this posting is There's a forum bug! |
Log in 
